DevOps · AWS · Docker · CI/CD
Push to production,
automatically.
I’m Balaji. This site is the demonstration: a commit to main
is built, tested and deployed to AWS in under a minute — no manual
step, no key on the server, and a rollback that happens on its own if the
new version is unhealthy.
- Deployed by GitHub Actions
- Runs on EC2 t3.micro
- Served by Nginx in Docker
Selected work
Automated CI/CD deployment pipeline
Designed, built and operated solo
A push to main builds a Docker image, smoke-tests it against
the routes the site actually serves, publishes it to Amazon ECR under an
immutable commit-SHA tag, then replaces the container on EC2 through
AWS Systems Manager. There is no SSH key, no inbound SSH rule, and no
credential stored on the server. If the new container never answers
/healthz, the previous image is restored automatically and
the pipeline still fails — a recovery is not a pass.
- 40–55 scommit to live
- 1reachable port
- 0credentials on EC2
- Autorollback, verified
Read the case study →
Source on GitHub
More write-ups are on the way. Each one follows the same format as above:
what it does, what it cost, what broke.
What I work with
Delivering and running
- Docker — images, layers, multi-stage builds
- GitHub Actions — jobs, dependencies, gated deploys
- AWS — EC2, ECR, IAM, Systems Manager
- Nginx — static serving, container config
- Linux — Ubuntu, systemd, permissions, TCP checks
Foundations
- Git — branching, commit history as the release version
- Bash — scripts that fail loudly and clean up after themselves
- YAML — pipeline definitions read by machines, reviewed by humans
- Least privilege — every identity scoped to one resource
- Cost awareness — right-sized instances, lifecycle policies
What I would build next on this
- OIDC federation instead of a stored access key
- A second instance behind a load balancer, for zero-downtime swaps
- Terraform for the AWS baseline, so the infrastructure is reviewable
- Failure notifications that fire on rollback too
Education
-
B.Tech, Computer Science & Engineering (AI & ML)
· 2026